Objective:

To determine the incidence of Kaposi's sarcoma (KS) in a prospective longitudinal cohort of HIV-1-infected individuals before during and after the introduction of highly active antiretroviral therapy (HAART) and to compare the incidence of KS between specific HAART regimens.

Design:

Univariate and multivariate analysis of 8640 HIV-1-infected individuals.

Methods:

The protective effect of HAART regimens based on either protease inhibitors (PI) or non-nucleoside reverse transcriptase inhibitors (NNRTI) on the development of KS was examined in prospectively recorded data to determine whether treatments based on the two types of drug were comparable with regard to a reduction in the incidence of KS.

Results:

A total of 1204 patients with KS were identified. The incidence of KS decreased from 30/1000 patient-years prior to 1995 to 0.03/1000 patient-years in 2001. Multivariate analysis showed that age, nadir CD4 cell count and antiretroviral class exposure were significant independent predictors of KS. NNRTI-based HAART (adjusted rate ratio, 0.42; 95% confidence interval 0.24–0.37) had a similar protective effect to PI-based HAART (adjusted rate ratio, 0.47; 95% confidence interval 0.38–0.58). Most patients who develop KS on HAART [30/35 (86%)] had evidence of virological treatment failure.

Conclusion:

PI-and NNRTI-based HAART regimens are equally effective as protection against KS. This is the first study to demonstrate a decreased incidence of an AIDS-defining disease with NNRTI-based therapy.